The long range objective of this project is to characterize images of cells from human breast aspiration biopsies, and to distinguish malignant from benign cells. Using the PEEP-DECIDE-GRAPH system, cell images were displayed and thresholded, and eighteen numeric characterizing features were extracted for each of approximately 100 cell nuclei. Using quadratic discrimination, combinations of as few as four features and in many cases just one feature allowed near perfect discrimination between cell categories. The project will continue along lines that would lead to a grading system for breast cells.